#asahi-dev · 2022-01-29 15:36
kettenisthe tas5770 speaker amplifier chip has a mode to mix the left and right channels into a single mono output#asahi-dev · 2022-01-29 15:35
kettenisuh?#asahi-dev · 2022-01-29 15:33
kettenispovik: there is another "REG_SERDES_CHANMASK" register at offset 0x18 in the serdes block and if I poke that one I can manage to get data in TDM slot 1 and have "stereo" output on the speaker of the mini#asahi-dev · 2022-01-28 19:29
kettenisotherwise it doesn't respond#asahi-dev · 2022-01-28 19:29
kettenisyeah, you defenitely need a 1ms after powering the tas5770 chip up#asahi-dev · 2022-01-27 12:15
ketteniscan't you just reserver the iova's used by the framebuffer to stop the iommu code fro messing about with those mappings?#asahi-dev · 2022-01-26 10:13
kettenisin the appropriate Kconfig file?#asahi-dev · 2022-01-26 10:12
kettenisdefault y if ARCH_APPLE#asahi-dev · 2022-01-26 10:12
kettenisisn't it just a matter of something like:#asahi-dev · 2022-01-26 10:11
kettenishow do other SoCs make sure their clock/reset/powerdomain drivers are enabled?#asahi-dev · 2022-01-26 08:48
kettenisalthough with the minimal hardware support that 5.16 and 5.17 have you may actually get away with that#asahi-dev · 2022-01-26 08:47
ketteniswon't work, since you'd miss the modifications that m1n1 makes#asahi-dev · 2022-01-25 21:48
kettenisI still need to figure out some of the details#asahi-dev · 2022-01-25 21:48
ketteniscorrect#asahi-dev · 2022-01-25 21:29
kettenisyay, I have working sound on the mini#asahi-dev · 2022-01-24 19:16
kettenisjust waiting for you slackers to catch up with me!#asahi-dev · 2022-01-24 19:16
kettenismini is defenitely feature complete for me (but then I have nvme ;))#asahi-dev · 2022-01-24 18:44
kettenisthere ;)#asahi-dev · 2022-01-24 18:44
ketteniss/nobody/only maz/#asahi-dev · 2022-01-24 18:39
kettenisto be honest, nobody is actually going to run 5.16 for real on these machines#asahi-dev · 2022-01-24 18:14
kettenismaz: not much that can be done about that I fear#asahi-dev · 2022-01-24 17:16
kettenisis it unhappy about adding the power domains?#asahi-dev · 2022-01-23 18:25
kettenisalyssa: what does "or not" mean? write-back cachable?#asahi-dev · 2022-01-23 18:25
kettenissince the framebuffer is in RAM#asahi-dev · 2022-01-23 18:24
kettenisbut in our case Normal NC is almost certainly the appropriate memory attribute#asahi-dev · 2022-01-23 18:23
ketteniswrite-combine is an x86 term that isn't very well defined#asahi-dev · 2022-01-23 13:00
kettenisif you get it wrong, you might not immediately see what the other side writes into the buffer#asahi-dev · 2022-01-23 12:58
kettenisthe SoC tends to be completely cache-coherent, so Normal WB is plausible#asahi-dev · 2022-01-23 12:51
kettenisyeah, SMC has to use nGnRnE#asahi-dev · 2022-01-23 12:39
kettenisdo we know what memory attributes macOS uses for these buffers?#asahi-dev · 2022-01-23 12:37
kettenisin particular whether it is mapped cachable or non-cachable#asahi-dev · 2022-01-23 12:36
ketteniswhat matters is what memeory attributes the other side uses#asahi-dev · 2022-01-22 11:50
kettenisso if audio-graph-card is a better fit you should probably use that#asahi-dev · 2022-01-22 11:49
ketteniswe have support for both in OpenBSD#asahi-dev · 2022-01-22 11:49
kettenisdon't remember saying that ;)#asahi-dev · 2022-01-21 11:12
kettenisbit of a hack, but maybe good enough to make progress#asahi-dev · 2022-01-21 11:11
kettenisthe first one that gets probed will claim the gpio and power the chip on#asahi-dev · 2022-01-21 11:10
kettenisyou don't really need the shared think I think#asahi-dev · 2022-01-21 11:10
kettenisthat was my thinking#asahi-dev · 2022-01-21 11:10
kettenisindeed#asahi-dev · 2022-01-21 11:09
kettenisfirst one to suspend will suspend them all yes#asahi-dev · 2022-01-21 11:08
kettenisor is the problem that gpios are exclusive?#asahi-dev · 2022-01-21 11:07
ketteniswith shutdown-gpio the first chip that gets probed would power them all up#asahi-dev · 2022-01-21 11:04
kettenisI suspect that using shutdown-gpio leads to less power consumption when suspended#asahi-dev · 2022-01-21 10:58
kettenisgood question#asahi-dev · 2022-01-21 10:56
kettenisand the driver takes sdz_gpio from "shutdown-gpio"#asahi-dev · 2022-01-21 10:55
kettenisright#asahi-dev · 2022-01-21 10:51
ketteniswhat happens if you use "shutdown-gpio" instead of "reset-gpio"?#asahi-dev · 2022-01-21 10:39
kettenisif the order in which the codecs get initialized is predictable, you could get away with only providing the reset gpio for the first one#asahi-dev · 2022-01-21 10:36
kettenispovik: ah, looks like I missed that because it is at the very top ;)